From: Tvrtko Ursulin tvrtko.ursulin@intel.com
There is this long standing nit of igt/tools/intel_error_decode asserting when you feed it an error state from a GPU the local libdrm does not know of.
To fix this I need a tweak in drm_intel_decode_context_alloc to make it not assert but just return NULL (which seems an already possible return value).
Signed-off-by: Tvrtko Ursulin tvrtko.ursulin@intel.com --- intel/intel_decode.c | 39 +++++++++++++++++++++------------------ 1 file changed, 21 insertions(+), 18 deletions(-)
diff --git a/intel/intel_decode.c b/intel/intel_decode.c index e0a516644314..be6f77984d65 100644 --- a/intel/intel_decode.c +++ b/intel/intel_decode.c @@ -3815,32 +3815,35 @@ drm_public struct drm_intel_decode * drm_intel_decode_context_alloc(uint32_t devid) { struct drm_intel_decode *ctx; + int gen = 0;
- ctx = calloc(1, sizeof(struct drm_intel_decode)); - if (!ctx) - return NULL; - - ctx->devid = devid; - ctx->out = stdout; - - if (intel_get_genx(devid, &ctx->gen)) + if (intel_get_genx(devid, &gen)) ; else if (IS_GEN8(devid)) - ctx->gen = 8; + gen = 8; else if (IS_GEN7(devid)) - ctx->gen = 7; + gen = 7; else if (IS_GEN6(devid)) - ctx->gen = 6; + gen = 6; else if (IS_GEN5(devid)) - ctx->gen = 5; + gen = 5; else if (IS_GEN4(devid)) - ctx->gen = 4; + gen = 4; else if (IS_9XX(devid)) - ctx->gen = 3; - else { - assert(IS_GEN2(devid)); - ctx->gen = 2; - } + gen = 3; + else if (IS_GEN2(devid)) + gen = 2; + + if (!gen) + return NULL; + + ctx = calloc(1, sizeof(struct drm_intel_decode)); + if (!ctx) + return NULL; + + ctx->devid = devid; + ctx->gen = gen; + ctx->out = stdout;
return ctx; }
Quoting Tvrtko Ursulin (2020-11-18 16:36:01)
From: Tvrtko Ursulin tvrtko.ursulin@intel.com
There is this long standing nit of igt/tools/intel_error_decode asserting when you feed it an error state from a GPU the local libdrm does not know of.
To fix this I need a tweak in drm_intel_decode_context_alloc to make it not assert but just return NULL (which seems an already possible return value).
Signed-off-by: Tvrtko Ursulin tvrtko.ursulin@intel.com
Good riddance, Reviewed-by: Chris Wilson chris@chris-wilson.co.uk -Chris
On 18/11/2020 17:04, Chris Wilson wrote:
Quoting Tvrtko Ursulin (2020-11-18 16:36:01)
From: Tvrtko Ursulin tvrtko.ursulin@intel.com
There is this long standing nit of igt/tools/intel_error_decode asserting when you feed it an error state from a GPU the local libdrm does not know of.
To fix this I need a tweak in drm_intel_decode_context_alloc to make it not assert but just return NULL (which seems an already possible return value).
Signed-off-by: Tvrtko Ursulin tvrtko.ursulin@intel.com
Good riddance, Reviewed-by: Chris Wilson chris@chris-wilson.co.uk
Thanks, now how can push to drm and is there some testing to be triggered before, or after?
Regards,
Tvrtko
Quoting Tvrtko Ursulin (2020-11-19 13:42:07)
On 18/11/2020 17:04, Chris Wilson wrote:
Quoting Tvrtko Ursulin (2020-11-18 16:36:01)
From: Tvrtko Ursulin tvrtko.ursulin@intel.com
There is this long standing nit of igt/tools/intel_error_decode asserting when you feed it an error state from a GPU the local libdrm does not know of.
To fix this I need a tweak in drm_intel_decode_context_alloc to make it not assert but just return NULL (which seems an already possible return value).
Signed-off-by: Tvrtko Ursulin tvrtko.ursulin@intel.com
Good riddance, Reviewed-by: Chris Wilson chris@chris-wilson.co.uk
Thanks, now how can push to drm and is there some testing to be triggered before, or after?
cd intel; for i in tests/gen*.sh; do $i; done
But clearly I haven't built libdrm since automake was dropped. -Chris
On 19/11/2020 13:52, Chris Wilson wrote:
Quoting Tvrtko Ursulin (2020-11-19 13:42:07)
On 18/11/2020 17:04, Chris Wilson wrote:
Quoting Tvrtko Ursulin (2020-11-18 16:36:01)
From: Tvrtko Ursulin tvrtko.ursulin@intel.com
There is this long standing nit of igt/tools/intel_error_decode asserting when you feed it an error state from a GPU the local libdrm does not know of.
To fix this I need a tweak in drm_intel_decode_context_alloc to make it not assert but just return NULL (which seems an already possible return value).
Signed-off-by: Tvrtko Ursulin tvrtko.ursulin@intel.com
Good riddance, Reviewed-by: Chris Wilson chris@chris-wilson.co.uk
Thanks, now how can push to drm and is there some testing to be triggered before, or after?
cd intel; for i in tests/gen*.sh; do $i; done
But clearly I haven't built libdrm since automake was dropped.
Thanks, all good:
$ for t in ../../intel/tests/gen*.sh; do bash -x $t; done ++ echo ../../intel/tests/gen4-3d.batch.sh ++ sed 's|.sh$||' + TEST_FILENAME=../../intel/tests/gen4-3d.batch + ./test_decode ../../intel/tests/gen4-3d.batch + ret=0 + test 0 = 1 + exit 0 ++ echo ../../intel/tests/gen5-3d.batch.sh ++ sed 's|.sh$||' + TEST_FILENAME=../../intel/tests/gen5-3d.batch + ./test_decode ../../intel/tests/gen5-3d.batch + ret=0 + test 0 = 1 + exit 0 ++ echo ../../intel/tests/gen6-3d.batch.sh ++ sed 's|.sh$||' + TEST_FILENAME=../../intel/tests/gen6-3d.batch + ./test_decode ../../intel/tests/gen6-3d.batch + ret=0 + test 0 = 1 + exit 0 ++ echo ../../intel/tests/gen7-2d-copy.batch.sh ++ sed 's|.sh$||' + TEST_FILENAME=../../intel/tests/gen7-2d-copy.batch + ./test_decode ../../intel/tests/gen7-2d-copy.batch + ret=0 + test 0 = 1 + exit 0 ++ echo ../../intel/tests/gen7-3d.batch.sh ++ sed 's|.sh$||' + TEST_FILENAME=../../intel/tests/gen7-3d.batch + ./test_decode ../../intel/tests/gen7-3d.batch + ret=0 + test 0 = 1 + exit 0
Regards,
Tvrtko
+ a bunch of recent committers to libdrm
Guys, anyone okay to push this patch? I can resend if required.
Regards,
Tvrtko
On 19/11/2020 13:58, Tvrtko Ursulin wrote:
On 19/11/2020 13:52, Chris Wilson wrote:
Quoting Tvrtko Ursulin (2020-11-19 13:42:07)
On 18/11/2020 17:04, Chris Wilson wrote:
Quoting Tvrtko Ursulin (2020-11-18 16:36:01)
From: Tvrtko Ursulin tvrtko.ursulin@intel.com
There is this long standing nit of igt/tools/intel_error_decode asserting when you feed it an error state from a GPU the local libdrm does not know of.
To fix this I need a tweak in drm_intel_decode_context_alloc to make it not assert but just return NULL (which seems an already possible return value).
Signed-off-by: Tvrtko Ursulin tvrtko.ursulin@intel.com
Good riddance, Reviewed-by: Chris Wilson chris@chris-wilson.co.uk
Thanks, now how can push to drm and is there some testing to be triggered before, or after?
cd intel; for i in tests/gen*.sh; do $i; done
But clearly I haven't built libdrm since automake was dropped.
Thanks, all good:
$ for t in ../../intel/tests/gen*.sh; do bash -x $t; done ++ echo ../../intel/tests/gen4-3d.batch.sh ++ sed 's|.sh$||'
- TEST_FILENAME=../../intel/tests/gen4-3d.batch
- ./test_decode ../../intel/tests/gen4-3d.batch
- ret=0
- test 0 = 1
- exit 0
++ echo ../../intel/tests/gen5-3d.batch.sh ++ sed 's|.sh$||'
- TEST_FILENAME=../../intel/tests/gen5-3d.batch
- ./test_decode ../../intel/tests/gen5-3d.batch
- ret=0
- test 0 = 1
- exit 0
++ echo ../../intel/tests/gen6-3d.batch.sh ++ sed 's|.sh$||'
- TEST_FILENAME=../../intel/tests/gen6-3d.batch
- ./test_decode ../../intel/tests/gen6-3d.batch
- ret=0
- test 0 = 1
- exit 0
++ echo ../../intel/tests/gen7-2d-copy.batch.sh ++ sed 's|.sh$||'
- TEST_FILENAME=../../intel/tests/gen7-2d-copy.batch
- ./test_decode ../../intel/tests/gen7-2d-copy.batch
- ret=0
- test 0 = 1
- exit 0
++ echo ../../intel/tests/gen7-3d.batch.sh ++ sed 's|.sh$||'
- TEST_FILENAME=../../intel/tests/gen7-3d.batch
- ./test_decode ../../intel/tests/gen7-3d.batch
- ret=0
- test 0 = 1
- exit 0
Regards,
Tvrtko _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel
dri-devel@lists.freedesktop.org
-
Chris Wilson -
Tvrtko Ursulin